AU2020101683A4 - Fault detection, location, and prediction within an electricity power transmission and distribution networks - Google Patents

Abstract

This patent is aimed at proposing a new cost-effective technique based on the power line voltage current characteristics to predict and identify the location of various abnormal and fault events in real time. By measuring the currents and voltages at both ends of the line, a unique fingerprint for the line during normal operation can be identified. Any change in this fingerprint can be detected and analysed by a software installed in the control centre in real time to identify the location, type and level of the abnormal events or emerging faults.

Description

Editorial Note 2020101683 There is only nine pages of the description

1. Patent title Fault detection, location, and prediction within an electricity power transmission and distribution networks

1.2 Summary of the invention The current industry practice to identify fault location in power lines is based on visual inspection, travelling waves and line impedance measurement. Unfortunately, these techniques are only developed to detect the fault location upon its occurrence without the ability to predict abnormal events that usually precede major faults and issue a timely warning signal to avoid any potential consequences for power line failures including bushfire. Furthermore, the current fault locating techniques exhibit some drawbacks that limit their wide practical implementation. This includes cost, access to required data and low accuracy when employed for specific power line topologies. This patent is aimed at proposing a new cost-effective technique based on the power line voltage-current characteristics to predict and identify the location of various abnormal and fault events in real time. By measuring the currents and voltages at both ends of the line, a unique fingerprint for the line during normal operation can be identified. Any change in this fingerprint can be detected and analysed by a software installed in the control centre in real time to identify the location, type and level of the abnormal events or emerging faults. Robustness of the proposed technique is assessed through Matlab/Simulink simulation analysis conducted on various case studies including series capacitor compensated lines along with a practical case study.

1.3 Innovation: The innovation of this patent comprises the following: -Introducing a novel online approach to predict the likelihood of a conductor failure. The proposed technique relies on constructing a fingerprint signature for the power line based on easily accessible voltage-current data measured at both ends of the line. -Developing a digital image processing-based technique to extract various image features of the proposed line signature. Changes in the line features due to emerging faults or abnormal events are analysed and early warning signal is issued. -Developing an Al-software code to estimate the equivalent electric circuit parameters of the power line and correlate the changes in these parameters with the location and level of various disturbance events within the line. -Integrating the above codes in one comprehensive software model to automate and standardize the proposed approach. -The proposed method does not call for additional measuring equipment as it uses the existing measuring and protection devices installed at the ends of the line. In case of the absence of such metering devices, in particular in distribution lines close to vegetation areas, simple and cheap data acquisition hardware setup will be developed and tested. -The proposed technique outperforms the current conventional utility practice in terms of cost, time and accuracy. -The proposed technique is easy to implement in existing and new installations of transmission and distribution lines. At this stage no such technique exists worldwide and the pioneering product of this research will be of the interest of global researchers as well as power utilities. In particular in Australia in which bushfire due to power lines failures has occurred several times without introducing a reliable solution so far.

2. Background Power lines represent vital links in electrical transmission and distribution networks. As they are usually operating under harsh environmental conditions, power lines are subjected to abnormal events such as wind gusts, snow accumulation, lightning, birds' activities and falling trees [1, 2]. Such events may lead to conductor swinging followed by short circuit faults. While no technique has so far been developed to accurately identify such abnormal events and hence predict faults before their occurrence, several techniques have been developed to identify fault locations within power lines. These techniques include travelling-waves [3] and apparent impedance [4] methods. Visual aerial method may also be utilised, however due to the high cost and risk involved, this method is used every 3 to 5 years for line condition assessment [5]. Travelling wave method identifies fault location based on the speed of the surge signals that appear at the fault location and propagate over the line in both directions along with the time it reaches the measuring location [6]. The key challenge for travelling wave-based methods is that the initial transient must be captured using a very precise time tagging device for accurate fault locating. Furthermore, as travelling waves reflect at discontinuity points, identification of the correct signal is critical. This method requires high frequency sampling and is costly for distribution applications that usually comprise load taps which create reflection difficulties [7]. Other limitations include the bandwidth of transducers and the inability to identify location of faults close to the relaying and measuring points and near zero fault inception angles [8]. Moreover, the accuracy of travelling wave-based technique decreases significantly when there is a discontinuity of the line characteristic impedance e.g. in case of two connected lines of different characteristic impedances or hybrid line / cable connection. Travelling waves may also be generated during normal line operation as a result of switching operations and lightning strikes thus identification of the exact signal related to the fault becomes a challenging task. The accuracy of travelling wave method is also affected by the stiffness of the transmission network and the speed of the propagated wave that is always assumed to be equal to the speed of light. Implementation of the travelling wave method calls for expensive high frequency sampling fault locating device installed at one or two terminals of the line [8]. In the impedance-based methods, fault location is identified based on the ratio of the measured reactance to the total reactance of the power line [9]. A calculation error arises when the fault and source currents are not in phase. This error can be reduced by adopting some modified techniques such as Tagaki, Eriksson and double ended-based methods [10, 11]. However, impedance-based methods call for precise information about the source and line impedances and a significant error is introduced to the calculations when the fault impedance is not a constant value. Furthermore, when impedance-based methods are implemented on three terminals or tapped lines, inaccurate calculations are obtained especially in the case of series compensated and un-transposed lines [12]. The above mentioned two techniques can be implemented on single or double terminals data based methods. In case of double terminals techniques, communication protocol between the line terminals such as satellite global positioning system (GPS) and phase measurement units (PMU) at both terminals must be used [13, 14]. In general, accuracy of current fault locating techniques is affected by load impedance dynamics, fault impedance, mutual coupling between parallel lines, power quality, line configuration, line compensation and saturation of current and potential transformers.

3. Proposed Approach and Methodology: To overcome the limitations of the current fault locating techniques as detailed above, this patent application proposes a new technique based on identifying a unique voltage-current characteristic (fingerprint) for power lines. The proposed fingerprint can be constructed every power cycle (20 ms for 50 Hz networks) and compared to instantly identify any changes and generate early warning signal so that a proper action can be taken to avoid any potential consequences. Any fault or abnormal event including short circuit faults, falling trees, bird activities, insulation failure, conductor corrosion, annealing and swinging will change the line equivalent electric circuit parameters (resistance, capacitance and mutual/self-inductance) and hence alter the proposed fingerprint in a unique way. This will facilitate continuous monitoring of the power lines in real time. The work plan of this patent application involves 4-tasks as elaborated below.

Claim-: Proposing a new line signature along with DIP code The new proposed fingerprint is a relation between the voltage difference at both terminals of the line and the summation of the currents at both sides as shown in Fig. 1. In this figure, the currents and voltages at the sending (Is, Vs) and receiving (IR, VR) ends are to be measured at both using PMU or a hardware setup developed in this application. The measured data can be synchronized using GPS time reference at both ends. As shown in Fig. 1, the proposed locus of AV=Vs-VR and Is+IR represents an ellipse whose geometrical dimensions depend on the currents and voltages at both ends and hence the condition of the line [15]. It is worth mentioning that the ellipse may exhibit changes due to normal dynamic operation and transient events. However, these changes can be easily discriminated from the changes due to fault and abnormal events as presented in the author's preliminary investigation [16]. Digital image processing (DIP) techniques enhance the analysis of graphic plots such as the elliptic plot shown in Fig. 1 using electronic devices. To automate the fault identification process, the proposed graphical fingerprint is represented by a two-dimensional matrix [A]xxy consists of finite number of pixels. Each pixel on the proposed image is represented mathematically as a(xi, yi) with an image intensity |al at a spatial location (xi, yi) with respect to X-Y coordinates. Various image features such as geometrical, texture analysis and invariant moments can be extracted from the measured ellipse to increase the accuracy of the proposed technique. The general methodology of the proposed DIP-based analysis technique is shown in Fig. 2. The process starts by acquiring the required data from both terminals of each phase of the power line (Is, Vs, IR, VR). The measured data are to be sampled over one complete power cycle using a sampling time of 0.4 ms which is a typical sampling rate for the PMU. These sampled parameters are used to construct the proposed voltage-current characteristic of each line at every power frequency cycle. Image pre-processing is conducted to resize and adjust the image colour and extension type. Segmentation process divides the pre-processed image into several segments to identify the exact characteristics from any unwanted elements such as noise and background. The optimal multistage Canny edge detection algorithm is used to detect the pixels of the proposed line characteristics and exclude other unwanted elements. Following the above steps, various unique image features can be extracted and used to calculate some classification metrics such as city-block distance (CBD), root mean square (RMS) and image Euclidean distance (IED) at each power cycle. These classification metrics are to be compared with the corresponding ones calculated at a previous power cycle to identify any anomalies. In case of significant metrics deviation from one cycle to another, an abnormal event is to be reported and an alarmed signal is to be issued at the control centre. Claim-2: Asset management and analytical approach code To enhance the reliability of the proposed online monitoring technique, this task is aimed at developing a complementary software code to aid in quantifying and classifying abnormalities in power line based on the variations of its equivalent electrical circuit parameters. The equivalent electric circuit parameters of the power line can be estimated through the line ABCD constants that can be calculated from the sending and receiving end parameters at various operating points. The main advantage of this analytical approach is its reliance on physical line parameters which adds more confidence to the diagnosis resulted from the DIP-code developed in Task-1. The line parameters are calculated for each type/level/location of conductor's abnormality using Simplorer and Maxwell finite element analysis (FEA). The estimated equivalent circuit parameters at each abnormal and fault event are to be compared with the reference parameters of the healthy line to identify the event's location, level and type. For practical implementation of this technique, A-based techniques including genetic algorithm (GA) and particle swarm optimisation (PSO), is used to correctly identify the line equivalent circuit parameters based on the measured currents and voltages at both ends of the line. A fitness function J is proposed to estimate the optimum set of the line parameters based on minimising the error of the line ABCD constants estimated from the line parameters; series resistance and self-inductance (Rs , Ls), mutual inductance between lines (M) and shunt capacitance and conductance (Csh, Gsh) at each iteration step and the ABCD parameters calculated from the measured voltages and currents at both ends of the line as below: VS =AVR + BIR; Is = CVR + DIR (1) A D = cosh(yl); B = Zc sinh(yl); C = sinh(yl) (2) where, the line characteristic impedance Ze and the overall line propagation constant yl are calculated from: Zc = Gsh+jO)Csh ; yl = V(Rs +ftoLs)(Gsh +j(Csh). The methodology of the GA is shown by the blue colour in the flow chart of Fig. 3 and is briefly elaborated below. •Initialization: a random population for the 5 electrical parameters (R, L,,M, Ch, Gsh) is created. •Evaluation: based on the parameters identified in the above step, the estimated ABCD constants are calculated using Eqn. (2) and are compared with the actual values calculated from the measured parameters in Eqn. (1). •Updating: a new population is created by applying selection and reproduction operators (elitism, crossover and mutation) and the best-fitted individuals are copied to the newly created population. • Termination: the above updating process is repeated till a global minimum value for the objective function or a maximum number of iterations is reached. Unlike GA which modifies population from generation to generation, PSO keeps the same population and updates the position of the population particles which flies through a search space with a velocity that is updated by movement inertia, self-cognition and social interaction. During each generation individual particles are accelerated to reduce its distance from the global optimum position. Fig. 3 shows the flow chart of PSO methodology (in red colour) which is quite similar to the GA described above except for updating the parameters through updating particles velocity (V) and position (X) using (3) and (4), respectively. V =wV +c 1 r (P - X,)czr2 kn(Pkb -X) (4)

X%+,M + Vk+1(5 X +1 =- Xk (5) where k is the iteration number, ci and c2 are cognitive and social acceleration parameters, ri and r2 are random numbers in the range 0 and 1, Pi is the best-known position of a particle m, Ppb is the best-known position within the population and w is the particle inertia. All results obtained from the simulation and analytical tasks mentioned above are utilized to develop an Al-based code to correctly identify abnormalities on power lines in real time and provide a suitable asset management decision as shown in the flow chart of Fig. 4. Claim-3: Experimental Validation Laboratory measurements are conducted on a transmission line model represented by cascaded pi-networks comprising R, L and C components as shown in Fig. 5. Other series of experiments are conducted on a power line that is emulated using a single twisted-pair of standard CAT5e cable of a 100Q surge impedance and a propagation velocity of0.64c (Fig. 5) so that the ABCD constants of the line can be calculated exactly. The line is energised using a signal generator of internal impedance so that the proposed ellipse can be investigated under different sending end voltage levels and frequencies. Voltage and current at both ends are measured and stored using PC oscilloscope. Various line faults are applied at different locations with various values of fault impedance to assess the capability of the proposed technique to detect high impedance faults. Claim-4: Practical Validation For this series of practical measurements, a complete setup as shown in Fig. 6 is to be developed to acquire the voltage/current at both ends of the investigated line span through potential transformer (PT) and current transformer (CT). As the nominal voltage of the PT's secondary side is usually 1OOV and the normal data acquisition card is of1OV rated voltage, a mini voltage transformer with voltage rating 100/3.5V and a precision of 0.2% is to be employed to interface the PT with the data acquisition card (DAQ). Also, a mini current transformer 5A/2.5mA with an acquisition circuit is to be developed. The developed setup is to be installed at both sides of the line and measured data are to be synchronised using GPS. In real field measurements, it is expected that the current and voltage waveforms are not perfectly pure sinusoidal due to the significant background electromagnetic noise. Therefore, a Fast Fourier Transform (FFT) based digital filtering technique is to be employed to eliminate the high harmonic orders and noises from the measured signals and extract the fundamental power frequency (50 Hz) component that is used to plot smooth ellipses. A compensation technique to eliminate the loading effect on the measured ellipse is also developed by converting the measured data at any loading condition to a pre-defined standard line loading condition. A graphical user interface (GUI) using LabVIEW will be developed to facilitate online display of the instantaneous current and voltage of each phase along with the corresponding ellipse diagnosis using the developed software code. In the proposed diagnosis system, the measured data are firstly processed using FFT to perform digital filtering to the unwanted frequencies from the measured voltage and current signals. The processed data are normalized to eliminate the influence of the load variation after which the V-I characteristic for each phase is plotted and the corresponding ellipse parameters are extracted and analysed. As the LabVIEW has professional graphical user interface in its front panel, it will be employed to process the data and display the results in real time.

4. Benefit Australian transmission and distribution networks comprise over 800,000 circuit kilometres of overhead conductors in service which represents an asset conservatively valued over several billion dollars. These lines are operating under various harsh environmental and operating conditions that accelerate the aging process of the line. In spite of the technology revolution took place within the last few decades; condition monitoring approaches to this asset class have not substantially changed. Reliable and cost effective methods of predicting the likelihood of a conductor failure have not yet been developed. Current utility asset management practice is based on visual inspections by aerial means which involves significant cost and risk. Conductor failure is often catastrophic and may lead to shock, electrocution and fire initiation. Hence, there is an essential need for developing new cost effective online prediction techniques for the likelihood of conductor failure to mitigate the cost of conductor replacement while maintaining a minimum risk to the community. When faults take place on power lines in the vicinity of rural areas, bush fire hazards may arise. The worst bushfire in Australian history, black Saturday that claimed 173 human lives, injured 5000 people, destroyed more than 2000 houses and burned over 4500 km2 of land started after a power line fault [17]. This was also the main reason for the recent bushfire in Tathra that burned over 1,250 hectares and destroyed over 100 structures [18]. Unfortunately, at this stage there is no technique featured with timely fault prediction to avoid such potential catastrophic consequences. This patent is aimed at developing a novel cost-effective technique to precisely predict and identify the abnormal events and faults locations within power lines in real time. Outcomes of patent proposal will introduce a new and significant contribution to the power lines condition monitoring and asset management. Results will be of direct relevance to the community and utilities. With a growing pool of ageing assets, the results will be of extreme value to Australian transmission and primary distribution networks in their strategy to maintain asset, optimize planned replacement, and minimize the possibility of catastrophic failures including accidental deaths associated with conductor's failure. The outcome will also benefit power utilities, end user and industry sector in terms of increased grids' reliability and cost savings due to outages for long periods. At this stage no such technique exists worldwide. The patent is clearly underscores ultimate research goal "objective of solving real world and community problems" and it aligns well with the national strategic priorities "resilient urban, rural and regional infrastructure", and "improved prediction, identification, tracking, prevention and management of emerging local and regional health threats".

5. Feasibility With the old electricity infrastructure in Australia (more than 70 years old), asset efficiency and optimisation has been prioritised on the agenda of the Australian government that allocated a large proportion of approximately $10 billion that is invested in the electrical energy sector every year for network upgrade and asset replacement. The proposed technique is aimed at preventing catastrophic consequence, in particular bushfire, due to power line failures by predicting and identifying the location of abnormal events within the power lines in real time. The proposed technique utilizes the existing infrastructure and protection devices equipped with the power lines to acquire the current-voltage data from both ends. With the ever-reducing cost of high-speed microprocessors and GPS, applications of synchronized measurements are rapidly becoming widespread across the power grids and are utilized to synchronize the collected data from both ends of the line. For power lines, in particular distribution lines, without metering devices from both ends, a simple and cheap data acquisition system will be implemented. 6. Samples of Results and Description of the drawings To assess the robustness of the proposed technique in identifying fault location in power lines, a modified version of the single line diagram for system-2 of the IEEE second benchmark model for computer simulation of sub-synchronous resonance as shown in Fig. 7 is simulated using Matlab / Simulink. The two 50 Hz, 100 MVA generators at buses 1 and 2 are connected by two parallel 150km, 500kV transmission lines (TLs) of which one is compensated using a series capacitor (Xc). As single line to ground faults represent about 70% of power line faults, a single line (phase-A) to ground fault is simulated at different locations of TL1 (m in km) of the system under study. The proposed line characteristic is obtained by measuring the phase voltages and currents at bus-1 (Vs, Is) and bus-2 (VR, IR). The measured sinusoidal waveforms are sampled over one complete power cycle using a sampling time of0.4ms to plot the proposed locus (AV versus Is+IR) in per-unit (pu) of the nominal rated power and voltage (100 MVA, 500 kV) as shown in the various scenarios below. A. Linefingerprint and referencefeatures During normal operation of the network shown in Fig. 7, the collected data are used to plot the proposed fingerprint of the three phases of TL without and with the series capacitor (50% compensation ratio is considered). It can be seen from Fig. 8 that the proposed characteristics for the three phases are identical during normal operation. The increment of the current in case of compensated line is attributed to the decrease in the line overall reactance. The simulation results in Fig. 8 are manipulated using a developed DIP code to extract four geometrical dimensions (GF): area of the ellipse gi, ellipse minor and major axes lengths ( 92 and g3 ) and ellipse focus point (94) along with the angle 0 between the ellipse major axis and the x-axis as listed in Table I. These features are considered as reference data set for the line without and with compensation. Results of the compensated line reveal the advantage of the proposed technique over impedance based methods that exhibits low accuracy in such case. B. Single phase to groundfaultfor uncompensated line In this case study, a bolted line to ground (L-G) fault is simulated at different locations of phase-A of TL1 when no compensation is employed. The obtained elliptic plots at different fault locations are compared with the line fingerprint as shown in Fig. 9. It can be seen that when L-G fault takes place at the middle of the line (75 km), the proposed locus does not encounter significant rotation. However, the ellipse entire area decreases when compared with the healthy fingerprint. When the fault occurs at locations less than 75 km (measured from bus 2), the ellipse rotates in anticlockwise direction with significant increase in the entire area. Ellipse rotation and area are more pronounced when the fault becomes closer to bus-2. When the fault occurs at a location more than 75 kin, mirrored elliptic plots are attained i.e. the locus obtained due to a fault at 125 km is almost similar to the obtained locus due to a fault at 25 km but in opposite direction. The extracted physical dimensions of the obtained results along with the angle of rotation are listed in Table II which reveal that each fault location within the line will have a unique impact on the proposed line characteristic. These results show the advantage of the proposed technique in detectingfaults close to the ends of the line which is one of the main issues of the travelling wave-based method. C. Discontinuity in the lie characteristic impedance One of the main drawbacks of the travelling wave-basedfault locating technique is when there is a discontinuity in the line characteristic impedance. For instance, when an overhead line is connected to a substation transformer through a short cable to avoid ferroresonance and lightening impacts on substation power transformers. A cable also may be connected to overhead transmission line at some sections due to right of way issues, or nature of terrestrial. To investigate the effect of such hybrid connection on the proposed method, a feeder of 5 km length and characteristic impedance of 10% of the line characteristic impedance is connected between TL1 and bus-2 of the system under study. Results of this case study are obtained for an uncompensated line as shown in Fig. 10. Results reveal almost the same effect of the fault taking place in a line with no cable connected (Fig. 9). This feature gives advantage of the proposed technique over travelling wave-based method in such line-cable configuration. D. Detection of abnormal events One of the key features of the proposed technique over existing ones is its ability to detect abnormal events and high impedance faults. To assess this feature, L-G fault is applied at the middle of TL1 (m =75 km) through various fault impedances (1, 2.5, 4 and 6k). The proposed line characteristic is obtained at each fault impedances. Results are shown in Fig. 11 with image features extracted and listed in Table III along with the fault current at each fault impedance. Results show that, the proposed line fingerprint will be impacted even when the fault current is only 3A. While the impact of such high impedance faults on some parameters such as ellipse entire area and minor axis length when compared with the corresponding features of normal line is insignificant, the effect on the major axis length, focus point and angle of rotation is noticeable. This implies the key role this technique may play in detecting abnormal events in the system prior tofault occurrence so that a timely action can be taken to avoid catastrophic consequences. To highlight the key advantageous of the proposed technique, a brief comparison of the current fault locating techniques and the new technique proposed in this research proposal is listed in Table IV. References

[1] H. Liang, Y. Liu, G. Sheng and X. Jiang, "Reproduction Methodology for Single Phase-to-Ground Faults in Overhead Transmission Lines," IEEE Access, vol. 5, pp. 17403-17413, 2017.

[2] N. I. Elkalashy, T. A. Kawady, W. M. Khater and A. L Taalab, "Unsynchronized Fault-Location Technique for Double-Circuit Transmission Systems Independent of Line Parameters," IEEE Transactions on Power Delivery, vol. 31, no. 4, pp. 1591-1600, Aug. 2016.

[3] H. A. Abd El-Ghany, A. M. Azmy and A. M. Abeid, "A General Travelling-Wave Based Scheme for Locating Simultaneous Faults in Transmission Lines," IEEE Transactions on Power Delivery, Early Access doi: 10.1109/TPWRD.2019.2931178.

[4] S. Das, S. Santoso, A. Gaikwad and M. Patel, "Impedance-based fault location in transmission networks: theory and application," IEEE Access, vol. 2, pp. 537-557, 2014.

[5] Electric Power Research Institute, EPRI,"Future Inspection of Overhead Transmission Lines," Technical Update, pp. 1-88, May 2008.

[6] F. Magnago and A. Abur, "Fault location using wavelets," IEEE Transactions on Power Delivery, vol. 13, pp.1475-1480,1998.

[7] Lotfifard, Saeed, Mladen Kezunovic, and Mirrasoul J. Mousavi. 2013. "A Systematic Approach for Ranking Distribution Systems Fault Location Algorithms and Eliminating False Estimates." IEEE Transactions on Power Delivery Vol. 28, No. 1, pp. 285-293, January 2013.

[8] Bo, Z.Q., G. Weller, and M.A. Redfern, "Accurate Fault Location Technique for Distribution System Using Fault-Generated High-Frequency Transient Voltage Signals" IEE Proceedings - Generation, Transmission and Distribution Vol. 146, Issue 1, pp. 73-79, 1999.

[9] Das, Swagata, Surya Santoso, Anish Gaikwad, and Mahendra Patel, "Impedance-Based Fault Location in Transmission Networks: Theory and Application." IEEE Access Vol. 2, pp. 537-557, 2014.

[10] T. Takagi, Y. Yamakoshi, M. Yamaura, R. Kondow, and T. Matsushima, "Development of a New Type Fault Locator Using the One-Terminal Voltage and Current Data," IEEE Power Engineering Review, vol. PER-2, pp. 59-60, 1982.

[11] L. Eriksson, , M. Saha, G.D. Rockefeller, "An Accurate Fault Locator with Compensation for Apparent Reactance in the Fault Resistance Resulting From Remote-End Infeed", IEEE Transactions on Power Apparatus and Systems Vol. PAS-104, Issue 2, pp. 423-436, 1985.

[12] "IEEE Guide for Determining Fault Location on AC Transmission and Distribution Lines - Redline," IEEE Std C37.114-2014 (Revision of IEEE Std C37.114-2004) - Redline, pp. 1-128,2015.

[13] V. A. Stanojevid, G. Preston and V. Terzija, "Synchronised Measurements Based Algorithm for Long Transmission Line Fault Analysis," IEEE Transactions on Smart Grid, vol. 9, no. 5, pp. 4448-4457, Sept. 2018.

[14] Ning Kang, Yuan Liao, January 2012. "Double-Circuit Transmission Line Fault Location with the Availability of Limited Voltage Measurements", IEEE transactions on Power Delivery, Vol.27, No. 1, pp.325 336 January 2012.

[15] A. Abu-Siada and S. Islam, "A Novel On-Line Technique to detect Power Transformer Winding Faults", IEEE Transaction on Power Delivery, Vol. 27, No. 2, pp. 849-857, April 2012.

[16] A. Abu-Siada, Saif Mir, "A new on-line technique to identify fault location within long transmission lines", Engineering Failure Analysis, Vol 105, Nov 2019, pp. 52-64.

[17] http://www.nma.gov.au/online features/defining moments/featured/black-saturday-bushfires

[18] https://en.wikipedia.org/wiki/2018 Tathra bushfire About the Inventor: Abu-Siada has extensive experience in the research topic and recognised credentials at national and international levels. Most of the inventor's publications (over 250 publications) and supervised PhD theses (21 completed HDR theses) are relevant to this research topic. Most of the publications in this research area have been published in the most prestigious IEEE transactions; some examples include IEEE transactions on Sustainable Energy (IF 7.65), power Electronics (IF 7.15); Industrial Electronics (IF 7.5); Industrial Informatics (IF 7.37); Power Delivery (IF 4.4). The inventor has a patent on a similar technique for power transformer online condition monitoring which will benefit electricity utilities by paving the way for the adoption of cheaper and more efficient online condition monitoring techniques for critical high voltage assets. This patent has been highlighted in "The Australian" newspaper and filmed in "YouTube" and has attracted investors from USA, Singapore, Germany and South Africa who contacted Curtin commercialization to express their interest in this invention. The idea of this research proposal has been selected among the top 10 Science and Technology Innovators in Australia in the CSRIO 2018 ON Accelerate program. The research idea and concept of predicting bushfire due to incipient faults in power lines in real-time was pitched to investors and potential industry partners during the CSIRO demo night in Sydney in 2018. Omicron Electronics (Germany) and Doble Engineering (USA), international leading companies in condition monitoring equipment, expressed interest in this new technique. The idea was highlighted on Particle science media https://particle.scitech.org.au/tech/finding-faults-in-the electrical-grid/ and a website https://research.csiro.au/vion/ has been developed by CSRIO to involve key industry players and investors in practical trials. Based on the research grants awarded to the inventor, he has built a strong record of accomplishment in developing innovative techniques in the proposal's research field, which resulted in substantial industry connections, publications (including edited book on condition monitoring of power transformers with the Institute of Eng. & Tech. (IET) UK, formerly EE), filed patents (2) and invited talks (more than 45). The inventor has extensive collaboration with international organisations, especially in China as a result of the two fellowships he was granted by the Australian Academy of Technological Sciences and Engineering to initiate collaboration with Chinese researchers. This has resulted in several joint funded projects and publications, especially in the area of fault diagnosis and condition monitoring. The CI has also strong collaboration with the Institut Teknologi Bandung and the state electricity company (Perusahaan Listrik Negara, PLN) in Indonesia. In recognition to his work in the area of fault diagnosis, asset management and condition monitoring, the inventor has been elected to Chair the International steering committee of the Condition Monitoring & Diagnosis conference since Oct 2018. He has also been a subject editor (condition monitoring and fault diagnosis) for IET GTD.

Fig. 1. Graphical illustration of the proposed approach.

Fig. 2. Flow chart of the proposed DIP technique.

Start

Acquire sending Generate initial values and receiving for the Line parameters end currents (Rs, Ls, M, Csh, Gsh,) and voltages

Calculate the fitness function J

Evaluate fitness function J 2020101683

Yes Yes Terminate GA based Terminate PSO based on criteria condition on criteria condition

No No Selection, crossover and Update the velocity of the mutation particles

Replace the worst Update the position of the chromosome with the particles bette offspring Print optimal parameters

Fig. 3. Flow chart of GA and PSO

Fig. 4 Flow chart for the overall proposed technique.

Signal generator 50Ω Transmission line (300m) 2020101683

PC oscilloscope

100Ω

Fig. 5 Transmission line models used in the experimental testing.

PT Voltage Signal Data Power acquisition acquisition transformer circuit card CT Current

User Lissajou Digital Industry Data process interface curve data computer

∆U Signal Load normalization Lissajous figure filtering algorithm constructution I1

Fig. 6. A schematic for the practical validation setup.

Fig. 7. Single line diagram for the TL network under study. 2020101683

0.15 0.15 Phase A Phase A

Phase B Phase B 0.1 0.1 Phase C Phase C

0.05 0.05

(pu) (pu)

0

R 0 R

V -V V -V S S

-0.05 -0.05

-0.1 -0.1

-0.15 -2.5 -2 -1.5 -1 -0.5 0 0.5 1 1.5 2 2.5 -0.15 -1.2 -1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1 1.2 I +I (pu) I +I (pu) S R S R

Fig. 8. Fingerprint for the three phases of TL1 (a) without series compensation and (b) with 50% series compensation.

0.15

0.15 Healthy Healthy 25 km 0.1 25 km 50 km 0.1 50 km 75 km 75 km 0.05 100 km 100 km 125 km (pu)

0.05 125 km (pu)

0 R V -V

0 R

S V -V

-0.05 S

-0.05

-0.1

-0.1

-0.15 -25 -20 -15 -10 -5 0 5 10 15 20 25 -0.15 I +I (pu) -25 -20 -15 -10 -5 0 5 10 15 20 25 S R I +I (pu) S R

Fig. 9. L-G fault at different locations. Fig. 10. L-G fault at different locations for line-cable connection.

0.15

0.1

0.05 (pu) R

0 -V S V

-0.05 R=1 kΩ

R=2.5 kΩ

-0.1 R=4 kΩ

R=6 kΩ

-0.15 -1.5 -1 -0.5 0 0.5 1 1.5 I +I (pu) S R

Fig. 11 Impact of fault impedance on the proposed method.

Table I. Reference GF for the investigated line. Aug 2020 2020101683

Table II. GF for bolted L-G fault at different locations.

Table III Geometrical dimensions for high impedance faults. Feature Normal 1 kΩ 2.5 kΩ 4 kΩ 6 kΩ 𝑔𝑔1 0.47 0.56 0.51 0.49 0.48 𝑔𝑔2 0.29 0.28 0.29 0.29 0.29 𝑔𝑔3 2.05 2.59 2.24 2.16 2.11 𝑔𝑔4 1.01 1.29 1.11 1.07 1.05 θ 0.34 2.08 1.35 0.96 0.69 Fault current - 50 A 23 A 9A 3A

Table IV Comparison of existing fault locating techniques and the proposed technique. Method Impedance Travelling waves Proposed technique Data and -Line / source impedance, -Length of the line, -Voltage-current at both devices -Phase voltage and current, -Wave velocity, ends, required -Fault type. -Expensive advanced relays -PMU or protection -PMU or protection devices. with complex algorithms. devices otherwise, data acquisition circuit is to be used. Accuracy -Not applicable for extra HV -Low accuracy for three -Still in research stage, and lines, terminal lines, unbalanced -Preliminarily results show limitations -Inaccurate when used for systems, low impedance faults the capability of the tapped radial, series capacitor and stiff networks, technique to detect faults: compensated and three terminal -Accuracy is affected by line -close to both ends, lines, parameters and assumed -of high impedance, -Low accuracy for low and velocity of the propagated -in unbalanced, un- high impedance faults, wave, transposed and -Error due to reactance effect -Inaccurate for faults close to compensated lines. and zero-sequence mutual sending or receiving ends. coupling. Prediction -Not applicable -Not applicable -Applicable